Content
@
https://warpcast.com/~/channel/ai-art
0 reply
0 recast
0 reaction
Luka →{protocell:labs}← 🎩
@luka
Back in 2018, before we had diffusion models, we had BigGAN. You could interpolate between categories of objects, like between a sailing ship and a church dome in this example. I think adversarial training for image models will make a comeback, similar like reinforcement learning made a comeback with DeepSeek.
0 reply
0 recast
5 reactions
